• 전기학회논문지P
• /
• 제67권1호
• /
• pp.42-46
• /
• 2018

In this paper, we propose fast object detection method of the cars by applying YOLOv2(You Only Look Once version 2) and UAVs (Unmanned Aerial Vehicles) while on the highway. We operated Darknet, OpenCV, CUDA and Deep Learning Server(SDX-4185) for our simulation environment. YOLOv2 is recently developed fast object detection algorithm that can detect various scale objects as fast speed. YOLOv2 convolution network algorithm allows to calculate probability by one pass evaluation and predicts location of each cars, because object detection process has simple single network. In our result, we could find cars on the highway area as fast speed and we could apply to the real time.

• 제어로봇시스템학회논문지
• /
• 제7권6호
• /
• pp.494-501
• /
• 2001

In this paper an ANFIS(Adativo Neuro-Fuzzy Inference System)- based fault detection and diagnosis for a closed loop control system is proposed. The proposed diagnostic system contains two ANFIS. One is run as a parallel model within the model in closed loop control(MCL) and the other is run as a series-parallel model within the process in closed loop(PCL) for the generation of relevant symptoms for fault diagnosis. These symptoms are further processed by another classification logic with simple rules and neural network for process and controller fault diagnosis. Experimental results for a DC shunt motor control system illustrate the effectiveness of the proposed diagnostic scheme.

• 한국진공학회:학술대회논문집
• /
• 한국진공학회 2013년도 제45회 하계 정기학술대회 초록집
• /
• pp.277.1-277.1
• /
• 2013

Molecular diagnostics consists of three processes, which are a sample pretreatment, a nucleic acid amplification, and an amplicon detection. Among three components, sample pretreatment is an important process in that it can increase the limit of detection by purifying nucleic acid in biological sample from contaminants that may interfere with the downstream genetic analysis such as nucleic acid amplification and detection. To achieve point-of-care virus detection system, the sample pretreatment process needs to be simple, rapid, and automatic. However, the commercial RNA extraction kits such as Rneasy (Qiagen) or MagnaPure (Roche) kit are highly labor-intensive and time-consuming due to numerous manual steps, and so it is not adequate for the on-site sample preparation. Herein, we have developed a rotary microfluidic system to extract and purify the RNA without necessity of external mechanical syringe pumps to allow flow control using microfluidic technology. We designed three reservoirs for sample, washing buffer, and elution buffer which were connected with different dimensional microfluidic channels. By controlling RPM, we could dispense a RNA sample solution, a washing buffer, and an elution buffer successively, so that the RNA was captured in the sol-gel solid phase, purified, and eluted in the downstream. Such a novel rotary sample preparation system eliminates some complicated hardwares and human intervention providing the opportunity to construct a fully integrated genetic analysis microsystem.

• 한국컴퓨터그래픽스학회논문지
• /
• 제29권1호
• /
• pp.1-11
• /
• 2023

딥러닝 기반의 이미지 세그멘테이션은 차선 인식을 위해 널리 사용되는 접근 방식 중 하나로, 차선의 키포인트를 추출하기 위한 후처리 과정이 필요하다. 일반적으로 키포인트는 사용자가 지정한 임계값을 기준으로 추출한다. 하지만 최적의 임계값을 찾는 과정은 큰 노력을 요구하며, 데이터 세트(또는 이미지)마다 최적의 값이 다를 수 있다. 본 연구는 사용자의 직접 임계값 지정 대신, 대상의 이미지에 맞추어 적절한 임계값을 자동으로 설정하는 키포인트 추출 알고리즘을 제안한다. 본 논문의 키포인트 추출 알고리즘은 차선 영역과 배경의 명확한 구분을 위해 줄 단위 정규화를 사용한다. 그리고 커널 밀도 추정을 사용하여, 각 줄에서 각 차선의 키포인트를 추출한다. 제안하는 알고리즘은 TuSimple과 CULane 데이터 세트에 적용되었으며, 고정된 임계값 사용 대비 정확도 및 거리오차 측면에서 1.80%p와 17.27% 향상된 결과를 얻는 것을 확인하였다.

• 산업공학
• /
• 제22권2호
• /
• pp.126-134
• /
• 2009

As product quality and yield are essential factors in semiconductor manufacturing, monitoring the main manufacturing steps is a critical task. For the purpose, FDC(Fault detection and classification) is used for diagnosing fault states in the processes by monitoring data stream collected by equipment sensors. This paper proposes an FDC model based on decision tree which provides if-then classification rules for causal analysis of the processing results. Unlike previous decision tree approaches, we reflect the structural aspect of the data stream to FDC. For this, we segment the data stream into multiple subregions, define structural features for each subregion, and select the features which have high relevance to results of the process and low redundancy to other features. As the result, we can construct simple, but highly accurate FDC model. Experiments using the data stream collected from etching process show that the proposed method is able to classify normal/abnormal states with high accuracy.

• 대한임베디드공학회논문지
• /
• 제11권3호
• /
• pp.163-171
• /
• 2016

Lane detection is a widely researched topic. Although simple road detection is easily achieved by previous methods, lane detection becomes very difficult in several complex cases involving noisy edges. To address this, we use a Convolution neural network (CNN) for image enhancement. CNN is a deep learning method that has been very successfully applied in object detection and recognition. In this paper, we introduce a robust lane detection method based on a CNN combined with random sample consensus (RANSAC) algorithm. Initially, we calculate edges in an image using a hat shaped kernel, then we detect lanes using the CNN combined with the RANSAC. In the training process of the CNN, input data consists of edge images and target data is images that have real white color lanes on an otherwise black background. The CNN structure consists of 8 layers with 3 convolutional layers, 2 subsampling layers and multi-layer perceptron (MLP) of 3 fully-connected layers. Convolutional and subsampling layers are hierarchically arranged to form a deep structure. Our proposed lane detection algorithm successfully eliminates noise lines and was found to perform better than other formal line detection algorithms such as RANSAC

• 반도체디스플레이기술학회지
• /
• 제17권1호
• /
• pp.54-61
• /
• 2018

In this study, two aspects were analyzed about the robot for removal of explosive devices. First, the cost analyses were performed to provide a reasonable solution for the acquirement of the system. It is processed by an engineering estimate method and the process was consisted of two ways : a system development expense and a mass production unit price. In additions, the resultant cost analyses were compared between the cases excluding and including a mines detection system. As results, in the case of the acquirement of the robot system for removal of explosive devices, it is recommended that the performance by improving the mines detection ability should be considered preferentially rather than the cost because the material cost for the mines detection system is negligible compared to the whole system cost. Second, as a way for improving the system performance by the mine detection function, the carbon nanotube (CNT) based sensor technology was studied in terms of sensitivity and simple productivity with presenting its preliminary experimental results. The detection electrodes were formed by a photolithography method using a photosensitive CNT paste. As results, this method was shown as a scalable and expandable technology for the excellent mines detection sensors.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2007년도 하계종합학술대회 논문집
• /
• pp.329-330
• /
• 2007

This paper presents a simple and robust face detection algorithm that can be utilized to video summary. We firstly apply the Illumination-compensation process for reducing the effect of brightness on the image. And then, we analyze the face region based on color in the YCbCr space to obtain the skin color. Also, we try the morphological image processing called closing algorithm to improve the detection. Experimental results demonstrate the effectiveness of our face detection algorithm that leads to 96.7 % precision ratio on average.

• Computers and Concrete
• /
• 제27권5호
• /
• pp.447-455
• /
• 2021

Research on damage detection methods in structures began a few decades ago with the introduction of methods based on structural vibration frequencies, which, of course, continues to this day. The value of important structures, on the one hand, and the countless maintenance costs on the other hand, have led researchers to always try to identify more accurate methods to diagnose damage to structures in the early stages. Among these, one of the most important and widely used methods in damage detection is the use of time-frequency representations. By using time-frequency representations, it is possible to process signals simultaneously in the time and frequency domains. In this research, the Short-Time Fourier transform, a known time-frequency function, has been used to process signals and identify the system. Besides, a new damage index has been introduced to identify damages in concrete piers of bridges. The proposed method has relatively simple calculations. To evaluate the method, the finite element model of an existing concrete bridge was created using as-built details. Based on the results, the method identifies the damages with high accuracy.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2005년도 추계종합학술대회
• /
• pp.739-742
• /
• 2005

In this paper, an efficient detection algorithm for the flicker, which is caused by mismatching between light frequency and exposure time at CMOS image sensor (CIS), is proposed. The flicker detection can be implemented by specific hardware or complex signal processing logic. However it is difficult to implement on single chip image sensor, which has pixel, CDS, ADC, and ISP on a die, because of limited die area. Thus for the flicker detection, the simple algorithm and high accuracy should be achieved on single chip image sensor,. To satisfy these purposes, the proposed algorithm organizes only simple operation, which calculates the subtraction of horizontal luminance mean between continuous two frames. This algorithm was verified with MATLAB and Xilinx FPGA, and it is implemented with Magnachip 0.18 standard cell library. As a result, the accuracy is 95% in average on FPGA emulation and the consumed gate count is about 7,500 gates (@40MHz) for implementation using Magnachip 0.18 process.